Fair and Balanced? News Media Bias in the Photographic Coverage of the 2016 U.S. Presidential Election

Download working paper here

The 2016 presidential election has renewed discussions about the impartiality of the news media. Scholars have studied this issue extensively, investigating newspapers, television and online news, yet the basic question remains unsettled: Is the media biased? In this paper, I focus on nine popular news websites covering the 2016 presidential election campaign. I apply computer vision techniques to photos of the candidates and their supporters at campaign rallies, automatically identifying their emotions. I expect that co-partisan news outlets would portray members of their favored side as happier and thus more positively. An analysis of this data shows that no such media bias exists. While both Donald Trump himself as well as his supporters consistently appear less happy than Clinton and her followers, there do not appear to be any differences between liberal and conservative news sources in this regard. I argue that these findings are the result of a polarized and profit-driven media environment, where the desire to portray one’s side in a positive light is balanced out by the incentive to keep readers apprehensive by playing up the other side’s chances of winning.